A Lack Of Situational AwarenessOn Thursday, Axios reported that Situational Awareness had sold its entire public-equity portfolio to Citadel. The news came one day after the Financial Times reported that the fund had sought fresh capital following heavy losses in the rout in AI stocks.Former OpenAI researcher Leopold Aschenbrenner founded Situational Awareness after publishing his influential 2024 essay series of the same name, which argued that rapid AI advances could produce artificial general intelligence before the decade’s end. Backed by Stripe’s Collison brothers and AI investors Nat Friedman and Daniel Gross, the fund reportedly grew to about $20 billion.The irony in the name wasn’t lost on observers: a fund called Situational Awareness appears to have had a blind spot about its own leverage, liquidity, and crowded positioning. Aschenbrenner may still be right about AI’s trajectory, but a sound long-term thesis couldn’t prevent short-term financing pressure from dictating the exit.Its latest public filing offers a snapshot. As of March 31st, its two largest disclosed common-stock positions were Bloom Energy, worth $878.7 million, and SanDisk, worth $724.4 million; it also disclosed call positions in both. A quarterly SEC filing can’t tell us what the fund still owned when it sold its equity book.Both companies appeared in our June 5th hedged portfolios, making the difference in risk architecture especially instructive.The Same Stocks, HedgedThe Hedged Portfolio Method is, to oversimplify, a hedged momentum strategy. Portfolio Armor starts with its current top names—the securities it estimates will generate the highest returns over the next six months, net of hedging cost. The investor chooses a “threshold,” the largest decline he’s willing to risk over that period, and our system builds the portfolio.Because every security is hedged, the portfolio can concentrate in the names with the highest expected returns instead of diluting them through broad diversification. When one sector dominates the rankings, that lets us participate in more of its upside while strictly limiting the downside. On June 5th, at the start of the recent momentum massacre, our system created such a portfolio for an investor with $3 million who was unwilling to risk a drawdown of more than 20% over the next six months. Screen capture via the Portfolio Armor website. The potential return is a high-end estimate of performance over the next 6 months; the expected return adjusts that for hedging cost and historical ratio of actual returns to potential returns.SanDisk (SNDK 0.00%↑) was one of the leading memory and storage names in the AI stack. It had been the Russell 1000’s strongest stock during the first half of 2026, rising as much as 858%, before leading the retreat among the index’s 25 biggest first-half winners.Bloom Energy (BE 0.00%↑) was another marquee name among investors betting on the AI buildout, because data centers need enormous quantities of reliable power. Together, Bloom and SanDisk represented nearly $1 million of the portfolio’s $3 million starting value.By late July, calling the reversal a “correction” almost seemed inadequate. Goldman Sachs’s High-Beta Momentum basket was on pace for its worst month on record. In individual AI-stack names, it looked more like a momentum massacre.Goldman's high beta momentum basket on track for worst monthly performance in history, worse than 2000 and 2009 pic.twitter.com/LRZG4g9i4m